A Preliminary Assessment of Ecosystem Vulnerability to ... · Ecosystem Vulnerability to Climate Change 6 6 DGVM alone or combine it with one of the methods mentioned above (e.g.,

Ecosystem Vulnerability to Climate Change 1

1

A Preliminary Assessment of Ecosystem Vulnerability to Climate Change in Panama

Laura Tremblay-Boyer and Eric Ross Anderson

ENVR 451 – Research in Panama

McGill University and the Smithsonian Tropical Research Institution

CATHALAC

Tel: (507) 317-3200

Fax: (507) 317-3299

E-mail: [email protected]

Address: Ciudad del Saber, Edificio 801, Clayton, Panama


2

Table of Contents

II. Host Institution.....................................................................................................................7

III. Study Site ............................................................................................................................8

IV. Methods ...............................................................................................................................8

A. Calculating the index’s components: four component of ecosystem vulnerability to

climate change .......................................................................................................................8

1) Vulnerability to sea level rise – EVCC1........................................................................9

2) Ecosystem geometry (edge effect and irregularity) – EVCC2 ......................................11

3) Climatic “space” of ecosystems – EVCC3 ..................................................................14

4) Species sensitivity – EVCC4 .......................................................................................18

B. Constructing the index: combining four variables into one.......................................21

C. Displaying the index and associated products:...........................................................21

D. Methods for applying index to current situations: .....................................................22

1) Degree of Human Intervention ..................................................................................22

2) Protected areas..........................................................................................................22

3) Species richness and endemism.................................................................................23

V. Results and Discussion .......................................................................................................23

A. Four components of ecosystem vulnerability to climate change ...................................24

1) Vulnerability to sea level rise – EVCC1......................................................................24

2) Ecosystem geometry (edge effect and irregularity) – EVCC2 ......................................28

3) Climatic ‘space’ of ecosystems – EVCC3...................................................................32

4) Species sensitivity – EVCC4 .......................................................................................37

B. Overall EVCC index .......................................................................................................44

C. Applications of the index ................................................................................................49

1) Degree of human intervention........................................................................................49

2) Protected areas .............................................................................................................51

3) Species richness and endemism .....................................................................................54

VI. Conclusions .......................................................................................................................57

X. References ...........................................................................................................................59

Appendix 2. Endemic Species in Panama .................................................................................63

Appendix 3. Special Focus Maps..............................................................................................65

Appendix 4. Chronogram of Activities & Time Spent ..............................................................67

Appendix 5. Notes of Gratitude – Contact Information.............................................................70


3

I. Introduction

Human-induced climate change has shifted status from hypothesis to reality during the

last decade. Global temperature averages are on the rise: of the twelve warmest years since 1850,

eleven of them were between 1995 and 2006, and the (IPCC, 2007) has very high confidence this

is a result of anthropogenic activities. Future scenarios predict global temperature to rise between

1.8 ºC and 4.0 ºC in the next ninety years. Because climate is very much a spatially

heterogeneous variable, the actual effects of global warming at the regional scale are likely to be

much more pronounced. Amongst other things, climate variability, heat waves, droughts and

heavy precipitation events are likely to increase in intensity and frequency, and changes in ocean

currents are expected to have unpredictable consequences (IPCC, 2007).

Climate is one of the key features defining the nature of ecosystems (Hawkins et al.,

2003). Current global warming has already been shown to impact the Earth’s biota at all scales

of organization from species to ecosystems, and the rate of change is likely to accelerate in the

future (Walther et al., 2002), especially if the effects of climate change interact with other drivers

like land-use use change and biotic exchange (Sala et al., 2000). Consequences of global

warming on biotic communities include shifts in species distribution and phenology (Parmesan,

2006; Thomas et al., 2004), increased risk of pathogen outbreaks (Pounds et al., 2006) and

disturbance of predator-prey cycles (Frederiksen et al., 2006). Complex effects resulting from

species interactions have also been documented and will undoubtedly be a common occurrence

given the high number of interacting species forming ecosystems (Ducklow et al., 2007; Walther

et al., 2002). Therefore, as the perceived effects of climate change are very likely to increase in

the future (IPCC, 2007), there is a very tangible risk that there will be important repercussions on

the identity and organization of biological systems.


4

Panama is a neotropical country that is host to a great number of species. Like most

developing countries it suffers from various social and economic issues, and these in turn are

reflected in the state of the environment. For instance, five hundred and seventy-five species are

listed on the 2006 IUCN’s Red List are from Panama (IUCN, 2006). It is thus very likely that

climate change will interact with socio-economic pressures and increase the strain already

present on Panamanian ecosystems. As a signatory to the Convention on Biological Diversity

(CBD), Panama has agreed to take the necessary steps to conserve and protect national

biodiversity in a sustainable manner (Emilio Sempris, pers. comm.). Given its high percentage of

remaining forest cover – 62.4% of the original primary forest (Institute, 2007) – and the large

area of the country covered by protected areas, it is a conservation opportunity with great

potential. For policy-making purposes, it is essential to identify the ecosystems that are most

likely to be affected by climate change in Panama as well as understand how this will impact the

rate of biodiversity loss.

The purpose of our internship is to evaluate the vulnerability of Panamanian ecosystems

to climate change. One of the main challenges is to define what exactly is meant by

“vulnerability” and what its proper measurement is. The IPCC defines vulnerability in terms of

climate change as, “the degree to which a system is susceptible to, or unable to cope with,

adverse effects of climate change, including climate variability and extremes” (IPCC, 2001).

This definition introduces yet another fuzzy term, “susceptibility,” which in turn refers to

exposure, sensitivity and adaptive capacity (Metzger et al., 2005). In this study we apply this

concept of vulnerability to ecosystems, in that vulnerable ecosystems are more likely to suffer

biodiversity loss from climate change. We realize this is still a rather loose definition, and the


5

general lack of understanding of the consequences of climate change for ecosystems makes it

difficult to define straightforward quantifiable responses.

As a preliminary step to assess the vulnerability of biological systems to climate change,

we reviewed methodologies used in both the scientific and institutional literature. We found that

most studies focus on the effects of climate change on biodiversity at the species-level scale. For

instance, there are numerous studies that have modeled predicted shifts in species’ distribution to

infer extinction risk of individual species (e.g., (Thomas et al., 2004; Thuiller et al., 2005b).

However, the relevance of this in terms of species’ ecology is being debated (Hampe, 2004),

especially since the results are strongly influenced by assumptions about species dispersal

abilities (Ibanez et al., 2006; Pearson, 2006). Other methods encountered include field

experiments, game-theory population models, expert opinion and outcome-driven modelling and

scenarios (Sutherland, 2006).

It is one thing to predict the effects of climate change on a species but quite another to try

to predict the consequences on ecosystems, which are made of countless interacting biological

and physical factors. While the relationship between species and ecosystem functions is far from

being understood (Peterson et al., 1998), there is a general consensus that biodiversity loss

decreases the resilience of ecosystems to biotic and abiotic disturbances (Hooper et al., 2005). In

a policy-making framework, it is certainly more useful to be able to assess vulnerability at the

ecosystem scale; however, due to its complexity and inherent uncertainties, this is a task that few

have attempted. For example, (Scholze et al., 2006) have looked at the effects of global warming

on key ecosystem processes (change in forest cover/carbon storage, wildfire frequency, and

freshwater availability) for the world ecosystems by using a Dynamic Global Vegetation Model

(DGVM). Most studies focused on the ecosystem scale actually use either some version of a


6

DGVM alone or combine it with one of the methods mentioned above (e.g., (Schroter et al.,

2005; Thuiller et al., 2006). The short timeframe, the potential amplitude of the task and our

level of expertise prevent us from using this type of methodology. We thus decided to perform a

preliminary vulnerability assessment where we would analyze general characteristics of

ecosystems that are likely to increase their sensitivity to climate change.

The main conservation objective underlining this study is the preservation of all of

Panama ecosystem types in the future, both to minimize biodiversity loss and to maintain

ecosystem services to Panamanians. Using scientific literature, we identified four key

components that could increase ecosystem vulnerability to climate change: (1) sea level rise, (2)

ecosystem geometry (area and shape), (3) climatic “space” and (4) species sensitivity (see the

next section for a detailed description). These four variables will be created and combined using

Geographical Information Systems (GIS) in order to construct an index of ecosystem

vulnerability to climate change (EVCC) for the ecosystems of Panama. Because we are well

aware of the uncertainties associated with a number of ecosystem responses to climate change,

we will not produce a definitive measure but rather rank each ecosystem patch according to its

vulnerability, from our calculations, to climate change under each of the four components. The

EVCC will be a weighted sum of the ranks for each component. It can thus be considered more

of a relative measure: it does not tell how much vulnerability an ecosystem has per se but how

much vulnerability, compared to other ecosystems, it has. The index will be presented as a series

of maps. Additionally, we will look at the relationship between the EVCC, the degree of human

intervention in ecosystems, the current network of protected areas, the distribution of endemic

species, and overall species richness, as these are relevant to Panama’s involvement in the CBD.

While we realize that there are a lot of uncertainties associated with this index, both with the data


7

used and the theory behind it, we feel it will be useful as a preliminary assessment for policy-

making, which will strive to meet Panama’s CBD objectives and improve the robustness of the

protected areas network. This project also has potential to serve as a foundation for continued or

more sophisticated studies on ecosystems and climate change in Panama.

II. Host Institution

The Water Center for the Humid Tropics of Latin America and the Caribbean

(CATHALAC) located in the City of Knowledge, Republic of Panama, is the headquarters of the

Regional Visualization and Monitoring System. CATHALAC operates SERVIR which has a

test bed and rapid prototyping facility managed by the NASA Marshall Space Flight Center at

the National Space Science and Technology Center in Huntsville, Alabama. SERVIR addresses

the nine societal benefit areas of the Global Earth Observation System of Systems (GEOSS):

disasters, ecosystems, biodiversity, weather, water, climate, oceans, health, agriculture, and

energy. It provides tools and technology to monitor resources and environmental conditions by

utilizing satellite imagery and other data sources. Results are presented to various levels of

regional governments and are also made freely available to the general public.

This resource monitoring involves daily and weekly generation of data on current

conditions such as weather, climate, marine environments, natural disasters, and ecological

productivity. Most of the results are used for immediate decision-support; therefore, there has

not been a concentrated effort on performing in-depth analyses of trends, which is what this

particular study will attempt to do.


8

The supervisors for this project are Emil Cherrington, the Geographic Information

Systems and Natural Resource Specialist for SERVIR, and Roxana Segundo, the Development

and International Cooperation Officer for CATHALAC.

III. Study Site

SERVIR shares and analyzes satellite and other types of data for all of Mesoamerica, but

this study focuses on Panama and its biodiversity at the ecosystem level. See the Introduction for

a more specific description of the country in the context of our project.

IV. Methods

In compliance with McGill University’s research ethics statement (2007), appropriate

measures were taken to maintain the integrity of this project. No certification was necessary for

research subjects since all of the work was done at CATHALAC in the computer laboratory;

however, credit is given to those organizations and people who provided our data. Results are

not definitive since the vulnerability assessments are based on climate change scenarios and

projections and because this is only a preliminary study that explores how to evaluate

vulnerability.

A. Calculating the index’s components: four component of ecosystem vulnerability to

climate change

Each of the four components was compared to or is related to a map of ecosystems in

Panama, which had been classified according to vegetation and level of human intervention.

There were thirty-seven ecosystem types and 1303 ecosystem patches (e.g., there are five

“Broadleaf evergreen altimontane rainforest” patches) (Appendix 1). The methodology was to


9

compare all components individually to each of the 1303 ecosystem patches, but generalizations

have been made about the thirty-seven ecosystem types. Since the domain analyses involve

different data types, the overlay methods were different and are described below. The general

processes of analyzing vulnerability in terms of each domain are provided, but the specific

technicalities of GIS have been spared from this paper. Indices were created for each domain,

according to its number (EVCC1,2,3,4)—each EVCCx receiving a value 1-10 (or 0-9 in the case of

EVCC1). They were later combined to create an overall vulnerability index—EVCC.

1) Vulnerability to sea level rise – EVCC1

Land loss due to sea level rise is a direct consequence of climate warming and will likely

impact a number of coastal ecosystems (e.g. mangroves). A digital elevation model (DEM)

provided elevation values at a ninety-meter scale. Since the IPCC (2001) “business as usual”

scenario predicts a 0.6 meter rise in sea level by 2099, all areas from zero to one meter in

elevation were identified. To avoid impractical results, only areas within one kilometer of

coastline were included, which will be called the “coastal buffer zone.” All areas of zero and

one meter elevations are called “red-zones.” The ecosystem patches were then assigned to the

coastal buffer zone, giving each “coastal patch” an identity (Figure 1). The following calculation

was used to determine EVCC1′:

Where cp

rc

A

AR = ,

ep

cp

A

AREVCC ='1 , (Equation 1)

where R is the area of red-zones in all coastal patches of an ecosystem patch (Arc) divided

by the area of all the coastal patches in an ecosystem patch (Acp), and Aep is the area of an

ecosystem patch (Figure 1). It is obvious that Acp is cancelled out in this equation, but it is

retained because of its importance in a further analysis.


10

Figure 1 This example of an ecosystem patch along the Gulf of Chiriqui contains over thirty coastal patches (some

too small to see at this scale). Other ecosystem patches (in different colors) break up patch #832. The tiny green dots

(90m resolution) are marked as red-zones.

EVCC1′ is not the vulnerability index for sea level rise but rather the density of red-zones

in all coastal patches multiplied by the density of coastal patch area in ecosystem patch area.

This means that EVCC1′ is a normalized value of how much red-zone there is in an ecosystem

patch, given only the areas within one kilometer of the coast. In order to determine EVCC1 the

EVCC1′ values were separated using the Geometric Interval quantitative classification scheme in

ArcMap. This scheme was created especially for continuous data and since EVCC1′ is a type of

density, this is a pertinent way to assign vulnerability values to each of the 1303 ecosystem

patches. Values zero to nine were assigned according to these intervals. The rationale behind

not using one to ten is that there are ecosystem patches that should be classified with “zero

vulnerability” to sea level rise because of their geographical location. For the other three

components, ranks of one to ten were used.


11

2) Ecosystem geometry (edge effect and irregularity) – EVCC2

Some ecosystems are likely to be more vulnerable to climatic disturbances brought about

by global warming because of their area and/or shape alone. Larger areas are less susceptible to

ramifications of stochastic perturbations (Janzen, 1983). Also, the shape, more specifically the

ratio of the perimeter to the area of the core, is important because of the edge effect and because

ecotones—transition zones with environmental gradients—are more sensitive to external factors

(Murcia, 1995; Saunders et al., 1991; Shafer, 1999). A number of tools have been developed in

the literature on optimal natural reserve shape, and they have been used to estimate this

component of ecosystem vulnerability to climate change. Two measurements for ecosystem

patch geometry were used. The first was an edge to core ratio calculation, which compares areas

(EVCC2a). The second was an attempt to evaluate of the irregularity of the ecosystem patch

(EVCC2b). A technique similar to EVCC1 was used to create more “buffer zones,” but this time

buffers were created for each ecosystem patch—not just the coastline. Measured inwards from

the border of a patch, a 500 meter “edge” was created for each patch (Figure 2). To calculate

EVCC2a′ a simple density equation was used:

e

ca

A

AEVCC ='2 , (Equation 2)

where Ae is the area of the edge in an ecosystem patch and Ac is the area of the core (even

though the inverse of this equation would obviously give the edge to core ratio, but a frequent

problem of Ac being equal to zero occurred because many patches were simply too small to have

any core). This gave a high value for those ecosystem patches that had a relatively high amount

of core compared to edge. This would mean that a higher EVCC2a′ indicates a lower

vulnerability. To create EVCC2a the values of EVCC2a′ were separated into five groups using the


12

Natural Breaks (Jenks) classification scheme in ArcMap (low EVCC2a is high vulnerability).

Because the Geometric Interval did not distinguish relative differences, the Natural Breaks was

able to break the EVCC2a′ values according to where there were natural boundaries or separations

in the distribution.

Figure 2 These ecosystem patches north of Yaviza, Darien display the 500m-wide edge areas. Comparing edge area

to core area, one can see that #757 has a much higher core:edge ratio than #763 does.

Irregularity was evaluated by measuring the area and perimeter of an ecosystem patch,

converting it into a perfect circle (while retaining the area), and measuring the perimeter of this

circle (Figure 3). The theory behind this method is that a circle has the least perimeter per area

of any shape, which means there are fewer edges that are susceptible to change. After measuring

the area and perimeter of the ecosystem patch and its perfect circle, EVCC2b′ was calculated:


13

circle

ep

bP

PEVCC ='2 , (Equation 3)

where Pep is the perimeter of the ecosystem patch and Pcircle is the perimeter of its perfect

circle. For the same reasons as EVCC2a′ the vulnerability index for EVCC2b was determined by

the Natural Breaks (Jenks) scheme into five intervals.

Figure 3 (not to scale) Conceptual method for obtaining EVCC2b. This shows that the ecosystem patch’s perimeter

is over six times as long as a perfect circle with the same area.

EVCC2 was simply obtained by adding the two methods together:

ba EVCCEVCCEVCC 222 += , (Equation 4 )

which yielded vulnerability values from one to ten.


14

3) Climatic “space” of ecosystems – EVCC3

Ecosystems have evolved to fit specific temperatures and precipitation regimes during the

year. Moreover, some ecosystems exist in regions where there are naturally large variations in

climate within and between years. Each ecosystem can thus be said to be adapted to fit a climatic

“space” that accounts both for average climate (for e.g. average temperature and precipitation)

and climatic variability. Climate change will affect the climatic conditions under which specific

ecosystems occur, and vulnerability should increase the further ecosystems are taken away from

the conditions they have been historically occuring in. Moreover, ecosystems with higher annual

variations in climate should be more resilient to climate change. The third component of the

EVCC, the climatic “space” of ecosystems, uses historical climate data and predicted future

climate to estimate which ecosystem patches, given the predicted change in climate where they

occur and the variation of climate they have been historically exposed to, should be more

vulnerable by climate change. We use temperature and precipitation data in our evaluation of

EVCC3, since they have both been shown to be important determinants of vegetation type

(Hawkins et al., 2003).

The historical climate data was obtained from WORLDCLIM, an electronic database that

provides high-resolution interpolated climatic data (1 km2) for the world, averaged from 1960 to

1990 (Hijmans et al. 2005). The data includes both standard variables like mean temperature and

precipitation but also a set of derived bioclimatic variables, such as the precipitation of the driest

quarter, that are more relevant to biological systems. The historical data obtained from

WORLDCLIM were temperature and precipitation for the months of July through September as

well as a number of bioclimatic variables described below.


15

The predicted future climate data for Panama was obtained from CATHALAC’s climate change

model (CCCM). The model was run at a scale of 36 km2 from 2005 until 2099. The data used in

EVCC3 was monthly temperature and precipitation average for july through september of the

years 2025, 2050 and 2099. The analysis was focused on those months because of data

availability but also because focusing on one season of a region with a lot of inter-annual

variation in climate allows to capture more variation. In the case where data from other parts of

the year become available and a third party would want to run the EVCC with them, we would

suggest not averaging the data together but doing separate analyses for biologically relevant

periods of the year. Of course, the magnitude of change in one period is likely to be correlated

with that in another and this would have to be accounted for. Additionally, as for now CCM has

only been run with one carbon emission scenario “business-as-usual”. It would have been

preferable to evaluate and compare the results of the EVCC according to a number of scenarios,

but this caveat is balanced by the fact that CATHALAC’s climate change model is the one with

the highest resolution for that region. Low-resolution is often invoked as one of the main

problems in studies of ecological response to climate change. CATHALAC’s climate change

model will be run with more scenarios in the future and other parties will have the possibility to

use the framework we have set up for the “business-as-usual” to evalutate EVCC under different

scenarios.

The analysis itself was conducted at a resolution of 1 km2, using the WORLDCLIM data as a

template. As a result, each 36 km2 pixel of the CCM were divided in 36 1 km

2 pixels which were

assigned the climate values of the CCM pixel they were derived from. While this induced a bias

in the EVCC3 in terms of spatial auto-correlation of the up-scaled pixel, it allowed to include a

greater number of ecosystem patches in the analysis and still provided information about the


16

magnitude of variation in climate for each 1 km2 pixel since the WORLDCLIM data occurs at

that resolution. A map of the excluded ecosystem patches is included in the “Results and

discussion” section (Figure 7).

The Gower Metric was used to evaluate predicted change in climate in terms of temperature and

precipitation for each 1 km2 pixel and the values obtained at the pixel level were averaged at the

ecosystem patch level. The Gower Metric is a measure of dissimilarity between environmental

coordinates that is often used in ecology, for example in the DOMAIN model of species

distributions. The Gower Metric increases as difference in climate increases and as historical

range in climate decreases. In other words, a pixel is deemed more vulnerable if it is predicted to

have a higher change in temperature and/or precipitation with the ecosystem patch were it occurs

having been exposed on average over 1960 – 1990 to a smaller intra-annual range of temperature

and precipitation.

The following steps were followed to build EVCC3 (see figure 4):

a. Climatic data was compilated for each 1 km2 pixel: historical data – mean monthly tempeture

(°C) and precipitation (mm) for the months of July, August and September, mean temperature of

the warmest and coldest quarters (three months), mean precipitation of the wettest and driest

quarters; predicted data – mean monthly temperature and precipitation for July, August and

September 2025, 2050 and 2099.

b. For both historical and predicted climate, mean monthly temperature and precipitation were

averaged over the months of July through September.

c. For each ecosystem patch, the minimum mean temperature of the coldest quarter (Tmin), the

maximum mean temperature of the warmest quarter (Tmax), the minimum mean precipitation for

the driest quarter (Pmin) and the maximum mean precipitation of the wettest quarter were


17

identified (Pmax). These values were substracted to calculate the temperature and precipitation

range for each ecosystem patch.

d. The Gower Metric for temperature and precipitation was calculated for each pixel for 2025,

2050 and 2099 by using the historical and future climate data compiled in Step 1 and the

temperature and precipitation ranges of the ecosystem patch where the pixel occurs calculated in

Step 2. An example for temperature is given here:

Gower Metric for cell i = (temphistorical – tempfuture )/ temprange of ecosystem patch containing i

e. A Gower Metric for temperature and precipitation for 2025, 2050 and 2099 was derived for

each ecosystem patch by averaging the Gower Metric within each category (temperature and

precipitation) of each of the pixels that occur in the patch.

f. For each ecosystem patch, an aggregated Gower Metric for temperature and precipitation was

calculated.

GM ecosystem patch = 4* GM2025 + 2* GM2050 + GM2099 (Equation 5)

This pattern of weighting was chosen because, given all else is equal, an ecosystem where the

rate of climate change is higher should be more vulnerable. Also, uncertainty in the accuracy of

the predicted climate data increases with time.

g. The aggreagated Gower Metrics for temperature and precipitation were compared amongst all

ecosystem patches and ecosystem patches were ranked from 1 to 10 using a … classification

scheme. A score of 10 was assigned to the highest Gower Metrics.


18

h. EVCC3 for each ecosystem patch was obtained by summing the ranks of that ecosystem patch

for the aggregated Gower Metric of temperature and precipitation. Under EVCC3 thus, an

ecosystem patch with a score of 20 was the most vulnerable compared to other ecosystem

patches in terms of both temperature and precipitation. Note that in the way EVCC3 is built,

temperature and precipitation were given equal weight in terms of their effect on ecosystem

vulnerability. This weighting is flexible and could be adjusted in future runs of the EVCC if

deemed relevant.

4) Species sensitivity – EVCC4

A large number of studies have modelled shifts in species distribution in response to

climate change and have gradually uncovered many basic characteristics that could make species

more vulnerable. These include for example geographic extent and basic niche properties

(Thuiller et al., 2005a), thermal range (Jiguet et al., 2006) and life form (for plants) (Broenniman

et al., 2006). Here a geographic extent of representative groups of vertebrates as defined by

NatureServe’s Infonatura (2007) was used for mammals, birds and amphibians to estimate

species’ sensitivity to climate change in each ecosystem. (Thuiller et al., 2005a)s’ study (2005a)

was consulted as a theoretical foundation since they found that species with smaller ranges were

less likely to have suitable habitat available for them in the future. Thus, in this fourth domain

ecosystem types that have a higher number of species with small ranges will be deemed more

vulnerable to climate change.

Estimated ranges for mammals, amphibians and birds of Panama were obtained from Infonatura,

an electronic database that provides conservation and education resources on vertebrates of Latin

America and the Caribbean (InfoNatura, 2007). The resulting coverage of Panamanian vertebrate


19

diversity was high: geographic information on ranges was available for 245 out of the 246

mammals, 845 of the 929 birds (including migratory birds), 192 out of the 196 amphibians.

Steps followed to build EVCC4:

A. Separate analysis of species sensitivity were performed for each group (mammals, birds,

amphibians) because of their inherrent different in range sizes (birds tend to have much bigger

ranges then amphibians) (see figure 4).

1. Total distribution area of species were calculated, including where the species occurs

outside of Panama, and assigned a rank from 1 to ‘x’ according to a quantile

classification scheme where a score of x was given to the 10% smaller, and thus most

sensitive, ranges. ‘x’ was selected according to the number of species in the group to

allow a finer definition of range categories, with ‘x’ being the highest in birds and the

smallest in amphibians.

2. From the species distribution we derived for each ecosystem patch which species

were theoretically occuring there. The average of the ranks calculated in step 1 of all

species of each group that occur within each ecosystem patch was calculated. In other

words, the average of the ranks for all the birds occuring in a given ecosystem patch

was calculated, the average of the ranks for all the amphibians, etc.

3. The species density of each ecosystem patch was calculated by dividing the total

number of species from each group that occur in the patch by the area-rank of the

patch from 1 to 10, as determined by ranking all ecosystem patches area with a

natural breaks (Jenks) scheme.


20

4. For each species group, the average rank divided by the species density for each

ecosystem patch were ranked on a scale of 10 using a geometric classification

scheme. This second ranking was necessary so that the different number of classes

assigned to each group in the initial ranking (see step 1) would not bias the resulting

average species sensitivity per ecosystem patch. The ranks from all three groups were

then added for each ecosystem patch.

B. EVCC 4 was calculated by summing for each ecosystem patch the ranks from all three groups

obtained in the last step and ranking (one last time) the aggregated average ranks according to a

quantile classification scheme. A rank of 5 was awarded to the ecosystem patches that had, on

average, the mammals, birds and amphibians with the smallest ranges for their respective groups.

Figure 4: An outline of the main steps leading to the final calculation of EVCC4

1. Distributions are ranked according to

their area relative to that of other species

of their group

2. For each ecosystem patch and for each of the

three groups, the average of the ranks of the

species occuring in the patch is calculated

3. The resulting average rank for each group is

divided by species density and ranked against

those of the other ecosystem patches.

Distributions of all mammals, birds, and amphibians species

found in Panama


21

B. Constructing the index: combining four variables into one

For each ecosystem type appropriate weights were assigned to each of the four

components defined above. This is important because some factors are not as pertinent to climate

change or cannot be as reliable in either the raw data or the analyses (Table 1).

Table 1 How to rank EVCC components: Numbers in italics represent the more influential

category for each domain, which is the deciding factor in the case of ties.

Domain

Pertinence to

Climate

Change

Reliability of

Data

Overall

Weight Scale

EVCC1 (sea level rise) 3 4 3 0 – 9

EVCC2 (patch geometry) 1 2 2 1 – 15

EVCC3 (climatic ”space”) 4 3 4 1 – 20

EVCC4 (species

sensitivity) 2 1 1 1 – 5

Considering that climatic ‘space’ is at the heart of this study, it has received the highest

weight. Sea level rise follows in second, and ecosystem geometry is the third most important.

Species sensitivity has been given the lowest weight because of the uncertainties in the theory of

species range as a driver of ecosystem patch vulnerability. There is also the problem of relying

on this species presence/absence data, because there remain many parts of Panama that are quite

unexplored. Given these ranks, the EVCC value for all of the ecosystem patches (except some

due to resolution issues, see EVCC3 results) is calculated:

4321 EVCCEVCCEVCCEVCCEVCCtotal +++= . (Equation 6)

C. Displaying the index and associated products:

The EVCC index has been displayed on a simple map of Panama divided into ecosystem

patches. Darker colors generally indicate higher vulnerability; refer to the legend accompanying

each map. Similar maps are also available for each domain of ecosystem patch vulnerability.


22

D. Methods for applying index to current situations:

For discussion and application purposes, the EVCC index was overlain on other types of

data. Doing this allows viewers to pinpoint areas that are especially vulnerable from the lack of

protection and/or the intrinsic biotic vulnerability (due to the endemism), and this will be a useful

decision-support tool in regards to the Panama’s CBD objectives and its focus on protected areas

and endemic species. The degree of human intervention, protected areas, endemism, and species

richness are four applications.

1) Degree of human intervention

Degrees of intervention were assigned to each ecosystem patch (values from zero to nine), which

reflect their UNESCO description. Table 2 guided the method in valuation:

Table 2 Guide to Assigning Intervention Values

Ecosystem Patch Description Intervention Value

Populated places 9

Shrimp farming 8

Agroforestry 7

Productive system: <10% natural or spontaneous vegetation 6

Productive system: 10-50% natural or spontaneous vegetation 5

Natural system: moderately high intervention (lowlands) 4

Natural system: moderately high intervention (mountains) 3

Natural system: medium intervention 2

Natural system: low intervention 1

All others 0

Overall EVCC was compared to degree of intervention to see which domain is the cause for the

most concern.

2) Protected areas

The overall EVCC scores for each ecosystem patch were again compared to a map of the

anthropological reserves, biological corridor, biological reserve, forest reserves, hydrological

protection zones, national marine parks, national parks, national wildlife refuge, private reserve,


23

protected forests, protective zone, recreation area, and wildlife refuges of Panama provided by

CATHALAC.

3) Species richness and endemism

The InfoNatura (Nature Serve, 2007) database provided the names of all mammal, bird

and amphibian species occuring in Panama as well as those that are endemics. Sixteen

mammals, eight birds, and twenty-nine amphibians are endemic to the country (Appendix 2).

Panama was divided into a grid of 10km x 10km squares and species richness and endemism

were calculated for each.

V. Results and Discussion

The results are presented in a similar framework as the Methods section. Discussions on

each set of results are provided directly after each results section so more congruent ideas can be

deduced. Since the focus of this assessment was done on the ecosystem patch level, results have

been generated for parts A and B that contain EVCCs for 1303 patches, but they are not provided

in this paper for the vast amount of space they require. For a more general analysis, summarized

results are also provided at the ecosystem level, in which thirty-seven ecosystem types have been

described in terms of different vulnerability components. A discussion on the different

impressions people can get from these maps is given in part C, and implications of our analysis

of further applications are addressed. For all results, critiques of the methods are provided in

order to initiate thoughts on how improvements can be made. Also, a large table of minimums,

maximums, averages, and variances, of EVCC1,2,,3,4 and overall EVCC summarized for each

ecosystem type is presented in Table 3.


24

A. Four components of ecosystem vulnerability to climate change

1) Vulnerability to sea level rise – EVCC1

A glance at Table 3 shows average EVCC1 values on a scale of 0-14. All of the

ecosystem types with a rank of zero have no land that is zero to one meter in elevation that is on

the one kilometer-deep coastline. Many of these individual patches are landlocked. A six-way

tie for the most vulnerable ecosystem patches (in the maximum EVCC1 column) goes to patches

of the following ecosystem types: mangroves, semideciduos tropical forests in lowlands (all

three levels of human intervention), small islands, and populated places. Sixty-two patches have

the highest EVCC1 value of 14, fifty-four of them being small islands. Over half of these small

islands are in Bocas del Toro, a province that also contains the other most vulnerable ecosystem

types such as mangroves, various types of forests in lowlands, swampy or marshy forests, and

occasionally flooded rainforests. Bocas is also the location of the only populated place patch

with the highest possible EVCC1 value. As seen in this province, often these ecosystem types are

neighbors to one another, which Map 1 attempts to display. The Colón and Free Trade Zone area

is a mix of similar ecosystem types with high patches of vulnerability. An almost identical mix

of ecosystem types with patches of high vulnerability exists along the Gulf of Chiriquí.

In terms of average EVCC1 the two most vulnerable ecosystem types (broadleaf

evergreen rainforests dominated by palms and in swamps, and coastal vegetation growing on

very new soils) only have seven and five patches, respectively, and account for an extremely

small percentage of total land in Panama. This could be a cause for concern because of the rarity

of this ecosystem type. See Figure 5 for a histogram of these results. After mangroves and

islands, which have already been discussed, come the salt and shrimp production ecosystem

types. Again these patches account for a very small area of the country and are located in the


25

Arco Seco of the Azuero Peninsula. They are still worth keeping in mind for their importance in

the economies and food production for the country (discussed in Applications section).

The method of selecting land of only zero to one meter in elevation involved using a

DEM on the 90m scale. Unfortunately is has a ±15m margin of error and slight imperfections

when overlaying it with the ecosystem map, some results are not completely accurate. Still, the

process identified the many of those tiny patches that would be more affected by this kind of

discrepancy and often labeled them as highly vulnerable. Also, by selectively choosing land that

was one kilometer or closer to the coast, some areas that of zero to one meter high were

disregarded. It could have be that inland valleys go even below sea level (and are not necessarily

vulnerable to sea level rise) or that expansive river deltas were wrongfully erased. The largest

bulk of elevations zero to one meter high that were erased occurred in the Arco Seco of the

Azuero Peninsua (dominating the shrimp and salt production). Whether or not this area should be

classified as more vulnerable, it has already received attention for being ranked highly.

Map 1 of vulnerability to sea level rise well-illustrates the particular coastal regions that

are more vulnerable than inland areas. Since the color scheme was applied to ecosystem patch

and not simply which coasts are susceptible to flooding, there is the possibility of misconception

in this demonstration of vulnerability. Most noticeable is the second largest ecosystem patch

(Productive system with less than 10% natural or spontaneous vegetation) that covers a large

portion of the Azuero Peninsula. Because many of its boundaries are coasts, there is a high

chance that parts of it will be susceptible to sea level rise. This shows up in the fact that it was

given an EVCC1 value of 3. Had this patch been divided into smaller sub-patches, a definite

difference in perception would occur.


26

Figure 5 Distribution of average EVCC1 values for each ecosystem type.

0 1 2 3 4 5 6 7 8 9

2b. BSOT latifoliado submontano - BI

2c. BSOT latifoliado submontano - PI

3a. BSOT latifoliado montano

3b. BSOT latifoliado montano - BI

3c. BSOT latifoliado montano - PI

4a. BSOT latifoliado altimontano

4b. BSOT latifoliado altimontano - Med. I

5. BSOT latifoliado nuboso

6b. BSOT aluvial ocasionalmente inundado - Med. I

6c. BSOT aluvial ocasionalmente inundado - domin. por Prioria

14. Plantaciones forestales

17. Vegetación de páramo

18a. Pant. de cyperáceas con abund. acum. de mat. org.

20. Flujo de lava con escasa vegetación

23. Carrizales pant. y form. sim. domin. por Typha

16. Sabanas arbol. de graminoides cortos inund.

1c. BSOT latifoliado de tierras bajas - PI

18b. Pantanos herbáceos salobres

8b. B. semideciduo tropical de tierras bajas - bastante intervenido

8a. B. semideciduo tropical de tierras bajas

10. B. deciduo por la sequia, latifoliado de tierras bajas

7a. BSOT pantanoso - domin. por dicotiledoneas

22. Albina con escasa vegetación

8c. B. semideciduo tropical de tierras bajas - poco intervenido

13a. Sist. prod. con veg. leñosa nat. o esp. signif. (10-50 %)

6a. BSOT aluvial ocasionalmente inundado

1a. BSOT latifoliado de tierras bajas

1b. BSOT latifoliado de tierras bajas - BI

12. Poblados

13b. Sist. prod. con veg. leñosa nat. o esp. signif. (<10 %)

7c. BSOT pantanoso - domin. por Campnosperma

15. Sist. prod. acuático-camaronera y salina

11. Isla menores de 140 ha

9. Manglares

21. Veg. costera de trans. sobre suelos marinos muy rec.

7b. BSOT pantanoso - domin. por palmas

EVCC1 rank per ecosystem type


27

Map 1


28

2) Ecosystem geometry (edge effect and irregularity) – EVCC2

It should first be pointed out that the distribution for the average EVCC2 (on a scale of 1

to 10) is not as widely spread as EVCC1. Despite this, ranges still exist at the extremes—EVCC2

of two to ten. Probably that EVCC2 is the sum of two sub-EVCC2 values with often opposite

values is because the average range is very concentrated from five to eight. Also, a value of one

is not possible because the minimum values for EVCC2a and EVCC2b were both one. Adding

two of the least vulnerable sub-values would mean that ‘2’ is the lowest possible EVCC2.

Gathering maximum values from Table 3, one can see that many of the coastal ecosystem

types are again highly vulnerable. This means that their shapes are of the most irregular

(probably long and thin in this case) and they are relatively small (having a large edge to core

ratio). A look at Map 2 shows that generally, ecosystem patches are smaller along the coastline,

likely because the change in land type can increase more dramatically from coast to inland than

from inland to inland. Agricultural systems also have some patches of very high vulnerability in

terms of geometry, but there is a caveat here. The geometry analysis meant to capture those

ecosystem patches with the most susceptibility to negative edge effects. Even though an

agricultural system could have an irregular shape, the theory behind its vulnerability does not

stand. That is to say, this geometrical analysis is much more pertinent for natural systems,

because human-altered systems are often very buffered from surrounding environmental changes

(later discussed in the Human Intervention section).

Ranking the ecosystem patches by highest EVCC2 (see for where to find complete

results) shows that nine out of ten patches are mangrove forests. It more is important to look at

ecosystem patch here because many ecosystem types have very few (and important to this

geometric analysis—small) patches. This is why the histogram (Figure 6) shows coastal


29

vegetation growing on very new soils and salt marshes with scarce vegetation to have the highest

EVCC2 ranks. They are definitely still important to consider because this type of ecosystem has

already been shown to be vulnerable to sea level rise.

The least vulnerable, which should be in theory the most resistant to surrounding

perturbations, include again many patches of lowland forest but also numerous patches in

mountainous regions and agricultural patches. The fact that lowland type ecosystems have

patches that have been classified as both least and most vulnerable further support the argument

that this geometry analysis is not good at summarizing vulnerability based on ecosystem type. It

is much more meaningful when considered for each patch.


30

Figure 6 Distribution of average EVCC2 values for each ecosystem type.

0 1 2 3 4 5 6 7 8 9








2a. BSOT latifoliado submontano




















12. Poblados






9. Manglares





31

Map 2


32

3) Climatic ‘space’ of ecosystems – EVCC3

EVCC3 is an indicator of how much climate will change on average for each ecosystem patch,

weighted by how much interannual variation in climate an ecosystem patch has been exposed to

during the last 40 years. Only 770 out of the 1303 ecosystem patches were evaluated because of

the resolution of the climate data and specificities of the type of spatial format (raster) that was

used for the analysis. Figure 7 shows highlighted in light blue the ecosystem patches that were

not included in the analysis. Their total area is less than 0.16% of the total country superficie so

that their exclusion from EVCC3 has minimal consequences on the overall EVCC.

Figure 7: Ecosystem patches (in light blue) that were not evaluated under EVCC3

A number of trends can be extracted by observing the EVCC3 maps (Map 3). First, the most

vulnerable regions of Panama, both in terms of temperature, precipitation and overall EVCC3,

are the Pacific and Carribean side of the west part of the country. The province of Bocas del

Toro shows the highest vulnerability overall. In general, the center of the country receives low

vulnerability rank, including the Peninsula de Azuero but excluding some ecosystem patches in

the Darien and around the Canal that are ranked with intermediate-high vulnerability. Given that

a great proportion of these systems are highly exploited for agriculture, this is a positive result in

economic terms. Isla Coiba has intermediate vulnerability, but its status as an island makes the

micro-climate less predictable. In general, bigger ecosystem patches are less vulnerable, making


33

++

EVCC3: Climate ‘space’ of ecosystems Map 3


34

an important proportion of the country’s superficie low vulnerability. The trend can be observed

at the ecosystem type level as well, as shown in figure 8. Figure 8 also shows that no ecosystem

type has on average very high vulnerability for both precipitation and temperature. Visually

comparing the temperature and precipiration maps shows that vulnerability in terms of

temperature and precipitation are well spatially correlated, except for the Canal area that has

more vulnerability in terms of precipitation and the Darien where vulnerability to temperature

change is much more important than the one to precipitation change. This trend also holds when

averaging to the ecosystem type level, as presented in figure Figure 9 which ranks ecosystem

types according to overall EVCC3 total but details the ranking (out of 10) for precipitation and

temperature. When a difference is observed, it is in general temperature that ranks higher, except

for mountain ecosystems where the opposite occurs. Figure 10 details the number of ecosystem

patches that have the same combination of rank for temperature and precipitation and shows that

the correlation holds only at smaller ranks. In other words, ecosystem patches that are extremely

vulnerable in terms of precipitation are generally not that vulnerable to temperature but

ecosystem patches that have moderate vulnerability often rank similarly in terms of temperature

and precipitation.


35

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

4b. BSOT latifoliado altimontano - Med. I – 0.09%

2c. BSOT latifoliado submontano - PI – 0.06%

3c. BSOT latifoliado montano - PI – 0.07%

4a. BSOT latifoliado altimontano – 0.52%

3b. BSOT latifoliado montano - BI – 0.61%

5. BSOT latifoliado nuboso – 0.29%

2a. BSOT latifoliado submontano – 9.12%

8c. B. semideciduo tropical de tierras bajas - poco intervenido – 0.79%

2b. BSOT latifoliado submontano - BI – 1.14%

22. Albina con escasa vegetación – 0.01%

10. B. deciduo por la sequia, latifoliado de tierras bajas – 0.09%

3a. BSOT latifoliado montano – 2.54%

8a. B. semideciduo tropical de tierras bajas – 9.51%

8b. B. semideciduo tropical de tierras bajas - bastante intervenido – 6.08%

13a. Sist. prod. con veg. leñosa nat. o esp. signif. (10-50 %) – 23.36%

15. Sist. prod. acuático-camaronera y salina – 0.1%

1b. BSOT latifoliado de tierras bajas - BI – 8.8%

18b. Pantanos herbáceos salobres – 0.03%

1a. BSOT latifoliado de tierras bajas – 14.35%

1c. BSOT latifoliado de tierras bajas - PI – 1.59%

7a. BSOT pantanoso - domin. por dicotiledoneas – 0.06%

16. Sabanas arbol. de graminoides cortos inund. – 0.06%

20. Flujo de lava con escasa vegetación – 0.08%

12. Poblados – 0.28%

23. Carrizales pant. y form. sim. domin. por Typha – 0.25%

13b. Sist. prod. con veg. leñosa nat. o esp. signif. (<10 %) – 15.34%

14. Plantaciones forestales – 0.07%

6b. BSOT aluvial ocasionalmente inundado - Med. I – 0.04%

11. Isla menores de 140 ha – 0.11%

9. Manglares – 2.86%

6c. BSOT aluvial ocasionalmente inundado - domin. por Prioria – 0.18%

17. Vegetación de páramo – 0.03%

21. Veg. costera de trans. sobre suelos marinos muy rec.– 0.05%

18a. Pant. de cyperáceas con abund. acum. de mat. org.– 0.02%

6a. BSOT aluvial ocasionalmente inundado – 1.04%

7b. BSOT pantanoso - domin. por palmas – 0.05%

7c. BSOT pantanoso - domin. por Campnosperma – 0.32%


Figure 8: Average EVCC3 per ecosystem type with proportion of Panama occupied by ecosystem


36

0 1 2 3 4 5 6 7 8 9 10

4b. BSOT latifoliado altimontano - Med. I – 0.09%

2c. BSOT latifoliado submontano - PI – 0.06%

3c. BSOT latifoliado montano - PI – 0.07%

4a. BSOT latifoliado altimontano – 0.52%

3b. BSOT latifoliado montano - BI – 0.61%

5. BSOT latifoliado nuboso – 0.29%

2a. BSOT latifoliado submontano – 9.12%

8c. B. semideciduo tropical de tierras bajas - poco intervenido – 0.79%

2b. BSOT latifoliado submontano - BI – 1.14%

22. Albina con escasa vegetación – 0.01%

10. B. deciduo por la sequia, latifoliado de tierras bajas – 0.09%

3a. BSOT latifoliado montano – 2.54%

8a. B. semideciduo tropical de tierras bajas – 9.51%

8b. B. semideciduo tropical de tierras bajas - bastante intervenido – 6.08%

13a. Sist. prod. con veg. leñosa nat. o esp. signif. (10-50 %) – 23.36%

15. Sist. prod. acuático-camaronera y salina – 0.1%

1b. BSOT latifoliado de tierras bajas - BI – 8.8%

18b. Pantanos herbáceos salobres – 0.03%

1a. BSOT latifoliado de tierras bajas – 14.35%

1c. BSOT latifoliado de tierras bajas - PI – 1.59%

7a. BSOT pantanoso - domin. por dicotiledoneas – 0.06%

16. Sabanas arbol. de graminoides cortos inund. – 0.06%

20. Flujo de lava con escasa vegetación – 0.08%

12. Poblados – 0.28%

23. Carrizales pant. y form. sim. domin. por Typha – 0.25%

13b. Sist. prod. con veg. leñosa nat. o esp. signif. (<10 %) – 15.34%

14. Plantaciones forestales – 0.07%

6b. BSOT aluvial ocasionalmente inundado - Med. I – 0.04%

11. Isla menores de 140 ha – 0.11%

9. Manglares – 2.86%

6c. BSOT aluvial ocasionalmente inundado - domin. por Prioria – 0.18%

17. Vegetación de páramo – 0.03%

21. Veg. costera de trans. sobre suelos marinos muy rec.– 0.05%

18a. Pant. de cyperáceas con abund. acum. de mat. org.– 0.02%

6a. BSOT aluvial ocasionalmente inundado – 1.04%

7b. BSOT pantanoso - domin. por palmas – 0.05%

7c. BSOT pantanoso - domin. por Campnosperma – 0.32%

Average Rank Precipitation Average Rank Temperature

Figure 9: Average rank for temperature and precipitation per ecosystem type and proportion of Panama

occupied by ecosystem


37

0

1

2

3

4

5

6

7

8

9

10

11

0 1 2 3 4 5 6 7 8 9 10 11

Ecosystem rank Temperature

Ecosystem

rank Precipitation

Figure 10: EVCC3 rank for temperature vs. precipitation, area of bubble represents the number of ecosystem

patches with that combination of rank

The trends gathered from EVCC3 show that the vulnerability to climate change in terms of the

climatic ‘space’ of ecosystems is asymmetrically distributed across the country and that it is

wrong to assume all ecosystems will react similarly. Such analysis is not only useful for

conservation purposes but also to increase the resilience of the agricultural system to global

warming.

4) Species sensitivity – EVCC4

EVCC4 is a measure of how many species sensititive to climate change an ecosystem patch

contains compared to other ecosystem patches, accounting for species density as a function of

species richness and area of the patch. EVCC4 was evaluated for all 1304 ecosystem patches. A

map of EVCC4 is presented in figure x, please refer to Appendix X for more detailed results.

Figure x shows average EVCC4 for each ecosystem type and figure x divides EVCC4 into its 3

components, birds, amphibians and mammals, per ecosystem type.


38

An interesthing trend is that mountains ecosystems appear more vulnerable. The north-west part

of the country is also more vulnerable, although this might be partly biased from the fact that the

biggest ecosystem patch was very species rich and that accounting for species density did not

completely eliminate the area effect. Islands generally got lower scores, this is probably due to

their low species density compared to the mainland. It would be interesthing to look more closely

at endemic species in islands as these are likely to be sensitive to climate change. It is

interesthing to visually compare the map of endemism in Panama (see page 56) with the map of

EVCC4. EVCC4 appears to be correlated with endemism, which would be expected given that

endemic species should have the smallest ranges of species within their groups. Finally, figure x2

tells us that species sensitivity ranks of mammals, amphibians and birds are not necessarily

correlated with themselves or with EVCC4, although mammals and birds are generally more

similar per ecosystem type.

The methodology used to measure EVCC4 appears appropriate. As shown in Table 2, accounting

for species density per ecosystem patch allowed to remove the area effect that occurs when one

only considers the average number of sensitive species. This effect resuts from the fact that a


39

EEVVCCCC44 –– SSppeecciieess sseennssiittiivviittyy

Map 4


40

larger area can contain relatively more species with smaller ranges. It would probably be

necessary to explore how the choice of the classification scheme (quantile and geometric) affects

the resulting EVCC4. It might prove worthwhile to justify a more scientifically grounded choice

of classification, that might be more appropriate to the type of data used and/or biologically

relevant. It is important to keep in mind that EVCC4 can only be as good as the initial species

distributions used to construct it so that care must be taken to have distributions as accurate as

possible to start with.

Due to data availability reasons, EVCC4 only looked at birds, mammals and amphibians. While

this yields interesthing results, uncertainty still remains as to how extinctions affect ecosystem

stability and what the relationship is between ecosystem diversity, stability and function (Hooper

et al., 2005). In other words, it is uncertain that a higher number of species that are sensitive to

climate change will make necessarily make an ecosystem more vulnerable to it. It is also

important to keep in mind that climate change is one of the many threats facing species and thus

EVCC4 cannot be considered in isolation for specific species. For instance, the puma is the

mammal with the widest distribution and as a result was not classified as sensitive. However, it is

extremely vulnerable to other drivers and already has disappeared from a number of ecosystems.

Ideally, EVCC4 would be built on actual distribution maps, not on potential, as what it aims at

ultimately is to estimate the risk of extinction and including species that have already

disappeared from the ecosystem add another bias to the measure. Additionally, a possible way to

make EVCC4 less uncertain in terms of ecosystem vulnerability and more relevant in terms of

the UNESCO’s ecosystem classification would be to look at tree distributions. Focusing on the


41

primary trophic level might also be more ecologically relevant when one is trying to link

sensitivity and risk of extinction to ecosystem stability.

The theory behind EVCC4 , while intuitive, is still new (Thuiller et al., 2005a) and is itself very

uncertain. It will be important to keep up-to-date with the scientific litterature to adjust EVCC4 in

terms of what is learned about species sensitivity to climate change. Despite its inherrent

uncertainty it still provides interesthing results about the distribution of species’ ranges while

being a first attempt to link studies of climate change at the species level with an understanding

of climate change vulnerability at the ecosystem level. The fact that EVCC4’s theory is rooted in

many outstanding issues about our understanding of ecosystems, including the diversity-stability

relationship, that it is easy to quantify and that it can be applied practically makes it a metric

worthwhile to investigate both for ecology and conservation.

Table 2: Average EVCC4 rank per ecosystem type

Average ecosystem patch area (m2)

Average species range EVCC4 Rank Average species range

and density

1 274771.59 2850058.63

2 11219409.43 55318330.35

3 21173321.81 104512348.92

4 39363662.95 174075080.23

5 244691013.67 16272026.14


42

0 1 2 3 4 5






9. Manglares









12. Poblados
























Figure 11: Distibution of EVCC4 values for each ecosystem type


43

0 1 2 3 4 5 6 7 8 9 10






9. Manglares









12. Poblados























Average Amphibian Rank Average Mammals Rank Average Birds Rank

Figure 12: Distribution of bird, mammal and amphibian sensitivity ranks for each ecosystem type, ranked by

total EVCC4


44

B. Overall EVCC index

Even though justification is provided in deciding how to weigh each of the EVCCx

values, the EVCC equation was still formulated somewhat arbitrarily. As Table 3 shows, each

individual EVCC was ranked on a different scale. Species sensitivity was on a scale from 1-5,

geometry from 1-10, sea level rise from 0-14, and climatic ‘space’ was ranked from 0-19.

This scoring scheme highlights the ecosystem types with both the highest single EVCC

values and average EVCC values. Red is the highest, orange the second, and yellow the third. In

the case of a tie, all values have been assigned the same color (e.g., Max_EVCC2 has one red

rank and a ten-way tie for orange, which means there will be no yellow). A column with “-no-F”

means that it does not consider any EVCC_F values even though they have been given colors.

The table is ranked by the overall score, which was determined by adding the number of each

color-highlight for ecosystem type, where red=3, orange=2, and yellow=1 (e.g., Mangroves: 3*(2

red) + 2*(0 orange) + 1*(2 yellow) = 8), but it does not include EVCC_F because there would be

redundancies in the score. Mangroves lead this ranking scheme because they had the highest

maximum EVCC twice. The highest average overall EVCC is 30 for tropical evergreen swampy

rainforests dominated by palms, but the variance of that value is extremely high. There are only

seven patches of this ecosystem type, so the possibility of variation evening out is much lower

than many other ecosystem types. But this argument does not work for islands, the most

numerous-patched ecosystem type, which also has a high overall average EVCC variance. This

can be explained by the large range that the islands carry—the minimum and maximum EVCCs

can be very high or very low, depending on the individual ecosystem patch.

An important general observation is that the ecosystems with the highest scores are

mangroves, new coastal vegetation, swamps or marshlands, lowlands, and small islands.


45

Table 3 Summary of individual (EVCC_1,2,3,4) and overall (EVCC_F) with statistics and rankings

Ecosystem Type

# patches

Min_EVCC1

Max_EVCC1

Ave_EVCC1

Var_EVCC1

Min_EVCC2

Max_EVCC2

Ave_EVCC2

Var_EVCC2

Min_EVCC3

Max_EVCC3

Ave_EVCC3

Var_EVCC3

Min_EVCC4

Max_EVCC4

Ave_EVCC4

Var_EVCC4

Min_EVCC_F

Max_EVCC_F

Ave_EVCC_F

Var_EVCC_F

Highlig

hts

Highlig

ht-n

o-F

RED-no-F

ORANGE-no-F

YELLOW-no-F

Score-no-F

I.A.5. Bosque de manglar 137 0 14 6.9 11.9 5 10 7.4 1.4 0 16 10.3 13.7 1 5 2.5 1.0 8 40 27.1 33.8 5 4 2 0 2 8

VI.B.3. Vegetación costera de transición sobre suelos marinos muy recientes 5 0 10 7.4 17.3 7 9 8.4 0.8 0 17 10.2 39.7 2 3 2.6 0.3 22 32 28.6 15.3 4 3 1 2 0 7

I.A.1.f.(2) Bosque siempreverde ombrofilo tropical aluvial ocasionalmente inundado 36 0 12 3.4 17.9 4 9 6.5 1.4 0 19 13.0 22.3 1 5 3.5 0.8 12 40 26.4 60.1 4 3 1 1 1 6

I.A.1.g.(2) Bosque siempreverde ombrofilo tropical pantanoso dominado por palmas 7 0 10 7.6 11.6 5 8 7.0 1.3 0 18 11.7 38.6 2 5 3.7 1.6 18 38 30.0 61.0 3 2 1 1 0 5

I.A.1.g.(3) Bosque siempreverde ombrofilo tropical pantanoso dominado por Campnosperma 8 0 8 4.6 10.6 5 8 6.1 1.0 0 18 14.0 40.9 2 5 3.3 0.8 9 35 28.0 69.7 3 2 1 1 0 5

I.A.3.a. Bosque semideciduo tropical de tierras bajas - bastante intervenido 76 0 14 2.2 16.8 4 9 6.2 1.6 0 15 8.3 8.2 1 5 2.9 1.4 10 30 19.6 25.5 2 2 1 1 0 5

I.A.3.a. Bosque semideciduo tropical de tierras bajas – poco intervenido 20 0 14 3.2 22.8 4 9 6.4 1.7 0 11 6.9 8.7 1 5 2.9 2.1 12 29 19.4 13.4 2 2 1 1 0 5

I.A.3.a. Bosque semideciduo tropical de tierras bajas 59 0 14 2.7 16.4 3 9 6.3 1.5 0 15 7.3 14.3 1 5 3.1 1.5 10 30 19.3 16.1 2 2 1 1 0 5

Isla menores de 140 hectareas 530 0 14 6.0 33.4 6 9 6.3 0.3 0 18 0.8 8.7 1 5 1.6 1.3 7 38 14.7 41.6 3 3 0 2 0 4

V.D.1.a. Pantanos de cyperáceas con abundante acumulación de material orgánico 3 0 0 0.0 0.0 6 7 6.7 0.3 11 18 13.3 16.3 3 5 4.3 1.3 23 27 24.3 5.3 2 2 0 2 0 4

SP.A. Sistema productivo con vegetación leñosa natural o espontánea significativa (10-50 %) 60 0 10 3.2 15.7 4 9 6.3 1.3 0 18 8.1 15.9 1 5 3.2 1.7 8 37 20.9 43.1 2 2 0 2 0 4

P. Poblados 21 0 14 3.9 27.8 5 8 6.8 0.7 0 15 8.6 20.8 1 5 3.1 2.0 16 40 22.4 39.5 2 1 1 0 0 3

I.A.1.d.(1) Bosque siempreverde ombrofilo tropical latifoliado altimontano (1500-2000 m Car, 1800-2300m Pac) - medianamente interv. 1 0 0 0.0 0.0 5 5 5.0 0.0 3 3 3.0 0.0 5 5 5.0 0.0 13 13 13.0 0.0 1 1 1 0 0 3

SP.B. Sistema productivo con vegetación leñosa natural o espontánea significativa (<10 %) 46 0 12 4.6 16.0 3 9 6.4 1.7 0 15 9.2 16.3 1 5 2.7 1.6 9 34 22.8 29.6 1 1 0 1 0 2

I.A.1.a.(1) Bosque siempreverde ombrofilo tropical latifoliado de tierras bajas - bastante intervenido 45 0 12 3.5 18.1 4 9 6.5 1.3 3 15 8.9 9.8 1 5 2.8 1.8 12 36 21.8 41.2 1 1 0 1 0 2

I.A.1.a.(1) Bosque siempreverde ombrofilo tropical latifoliado de tierras bajas 77 0 12 3.4 16.0 2 9 6.1 1.7 0 17 8.7 15.0 1 5 3.2 2.1 11 32 21.5 26.4 1 1 0 1 0 2

VI.D. Albina con escasa vegetación 1 3 3 3.0 0.0 8 8 8.0 0.0 8 8 8.0 0.0 2 2 2.0 0.0 21 21 21.0 0.0 1 1 0 1 0 2

I.A.1.c.(1) Bosque siempreverde ombrofilo tropical latifoliado montano (1000-1500m Caribe, 1200-1800 m Pacífico) - bastante intervenido 14 0 0 0.0 0.0 5 7 6.0 0.8 3 13 6.6 8.7 3 5 4.6 0.4 12 25 17.1 11.1 1 1 0 1 0 2


46

VI.A.d. Flujo de lava con escasa vegetación 4 0 0 0.0 0.0 5 7 5.8 0.9 6 14 9.8 18.9 4 5 4.5 0.3 16 23 20.0 12.7 1 1 0 0 1 1

I.A.1.d.(1) Bosque siempreverde ombrofilo tropical latifoliado altimontano (1500-2000 m Caribe, 1800-2300 m Pacífico) 4 0 0 0.0 0.0 5 8 6.0 2.0 3 12 6.5 19.0 4 5 4.5 0.3 13 22 17.0 15.3 1 1 0 0 1 1

I.A.1.b.(1) Bosque siempreverde ombrofilo tropical latifoliado submontano (500-1000 m Caribe, 700-1200 m Pacífico) - poco intervenido 2 0 0 0.0 0.0 4 6 5.0 2.0 5 6 5.5 0.5 4 5 4.5 0.5 13 17 15.0 8.0 1 1 0 0 1 1

I.A.1.g.(1) Bosque siempreverde ombrofilo tropical pantanoso dominado por dicotiledóneas 5 0 9 3.0 18.0 6 8 7.0 0.5 9 11 9.6 0.8 3 5 4.2 0.7 20 32 23.8 25.2 0 0 0 0 0 0

I.A.1.f.(2)(a) Bosque siempreverde ombrofilo tropical aluvial, ocasionalmente inundado dominado por Prioria 6 0 0 0.0 0.0 5 8 6.2 1.4 9 14 12.2 4.6 3 5 4.0 0.4 18 26 22.3 7.5 0 0 0 0 0 0

I.A.1.a.(1) Bosque siempreverde ombrofilo tropical latifoliado de tierras bajas - poco intervenido 32 0 12 1.7 11.1 4 8 6.3 1.4 5 16 9.4 11.0 1 5 3.8 2.1 15 32 21.2 26.1 0 0 0 0 0 0

V.C.2.b. Vegetación de páramo 2 0 0 0.0 0.0 5 6 5.5 0.5 12 13 12.5 0.5 3 3 3.0 0.0 21 21 21.0 0.0 0 0 0 0 0 0

SP.C. Sistema productivo acuático-camaronera y salina 6 0 12 4.7 21.9 5 8 6.8 1.4 0 11 7.3 14.3 2 2 2.0 0.0 19 25 20.8 5.8 0 0 0 0 0 0

V.E.1.a.2. Pantanos herbáceos salobres 4 0 8 2.0 16.0 6 8 7.0 0.7 7 12 9.3 4.3 2 4 2.5 1.0 17 31 20.8 46.9 0 0 0 0 0 0

I.B.1.a.(1) Bosque deciduo por la sequía, latifoliado de tierras bajas 5 0 8 2.8 15.2 4 7 5.8 1.7 6 10 8.2 2.2 3 5 3.8 0.7 17 29 20.6 24.3 0 0 0 0 0 0

SP.B.1. Plantaciones forestales 6 0 0 0.0 0.0 6 7 6.3 0.3 7 15 10.8 9.0 2 4 3.2 1.0 17 23 20.3 4.7 0 0 0 0 0 0

V.A.2.d. Sabanas arboladas de graminoides cortos inundables 5 0 6 1.2 7.2 6 8 7.0 0.5 7 12 9.6 3.3 2 4 2.4 0.8 19 23 20.2 3.2 0 0 0 0 0 0

VII.B. Carrizales pantanosos y formaciones similares principalmente de Typha 5 0 0 0.0 0.0 4 7 5.4 1.8 9 11 10.0 1.0 4 4 4.0 0.0 17 21 19.4 3.3 0 0 0 0 0 0

I.A.1.f.(2) Bosque siempreverde ombrofilo tropical aluvial ocasionalmente inundado - medianamente intervenido 1 0 0 0.0 0.0 5 5 5.0 0.0 11 11 11.0 0.0 2 2 2.0 0.0 18 18 18.0 0.0 0 0 0 0 0 0

I.A.1.c.(1) Bosque siempreverde ombrofilo tropical latifoliado montano (1000-1500 m Caribe, 1200-1800 m Pacífico) 18 0 0 0.0 0.0 4 8 5.6 1.3 3 14 8.2 14.9 2 5 3.9 1.3 10 24 17.8 16.8 0 0 0 0 0 0

I.A.1.e.(1) Bosque siempreverde ombrofilo tropical latifolado nuboso (2000-3000 m Caribe, 2300-3000 m Pacífico) 3 0 0 0.0 0.0 6 6 6.0 0.0 5 9 7.3 4.3 3 5 4.3 1.3 14 20 17.7 10.3 0 0 0 0 0 0

I.A.1.b.(1) Bosque siempreverde ombrofilo tropical latifoliado submontano (500-1000 m Caribe, 700-1200 m Pacífico) - bastante interv. 21 0 0 0.0 0.0 3 8 5.6 1.5 4 13 8.0 10.6 2 5 3.8 1.8 12 22 17.4 11.4 0 0 0 0 0 0

I.A.1.b.(1) Bosque siempreverde ombrofilo tropical latifoliado submontano (500-1000 m Caribe, 700-1200 m Pacífico) 27 0 0 0.0 0.0 3 8 5.7 1.4 2 14 7.6 11.2 2 5 3.5 1.1 9 24 16.9 13.5 0 0 0 0 0 0

I.A.1.c.(1) Bosque siempreverde ombrofilo tropical latifoliado montano (1000-1500 m Caribe, 1200-1800 m Pacífico) - poco intervenido 1 0 0 0.0 0.0 5 5 5.0 0.0 6 6 6.0 0.0 4 4 4.0 0.0 15 15 15.0 0.0 0 0 0 0 0 0


47

Figure 13 Distribution of average overall EVCC values for each ecosystem type

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32




























12. Poblados





9. Manglares




EVCC rank per ecosystem type


48

Map 5


49

C. Applications of the index

Only the overall EVCC values are compared to the three applications. Relationships

between the EVCC index and the applications are provided and discussed in each section.

1) Degree of human intervention

Before comparing the EVCC to this measurement, the percent area of each intervention

level was calculated (Table 4).

Table 4 Percentage of human intervention

Level of Intervention Total area of intervention type (ha)

% area of Panama

0 None documented 3096464.7364 41.58%

1 Natural system with low intervention 187260.6452 2.51%

2 Natural system with medium intervention 9389.7263 0.13%

3 Natural system with high intervention in mountains 130335.2176 1.75%

4 Natural system with high intervention in lowlands 1107875.1517 14.88%

5 Productive system with 10-50% natural veg 1739669.8057 23.36%

6 Productive system with <10% natural veg 1142648.6367 15.34%

7 Agroforestry 5267.1732 0.07%

8 Shrimp / Salt production 7391.1249 0.10%

9 Populated place 21170.0424 0.28%

All area of Panama 7447472.2601 100.00%

These results are based solely on UNESCO’s descriptions of ecosystem types and their

delineations on its ecosystem map. The three mid-levels (4, 5, and 6) of intervention account for

over half of the land in Panama. The areas with the highest levels of intervention (agroforestry,

shrimp and salt production, and populated places) cover less than one-half percent of the land,

but are should not be deemed insignificant because of the strong human interest placed in these

systems (e.g., ecosystem services).


50

Table 5 EVCC values compared to Degrees of Intervention

Level of intervention Avg_EVCC1 Ave_EVCC2 Ave_EVCC3 Ave_EVCC4 Ave_EVCC

0 none 5.0987 6.4086 4.6817 2.2658 18.4548

1

Natural system with low intervention 2.1636 6.2727 8.2545 3.4727 20.1636

2

Natural system with medium intervention 0.0000 5.0000 7.0000 3.5000 15.5000

3

Natural system with high intervention in mountains 0.0000 5.7714 7.4286 4.0857 17.2857

4

Natural system with high intervention in lowlands 2.7025 6.3223 8.5207 2.8678 20.4132

5

Productive system with 10-50% natural veg 3.2333 6.3167 8.1333 3.2333 20.9167

6

Productive system with <10% natural veg 4.5652 6.3696 9.1522 2.7391 22.8261

7 Agroforestry 0.0000 6.3333 10.8333 3.1667 20.3333

8 Shrimp / Salt production 4.6667 6.8333 7.3333 2.0000 20.8333

9 Populated place 3.9048 6.7619 8.5714 3.1429 22.3810

Red = highest average; orange = second highest; yellow = third highest.

Blue = most commonly ranked with high vulnerability

Table 5 demonstrates the individual and overall average EVCC values for each level of

intervention. Notable levels are zero, six, eight, and nine (highlighted in beige) because of the

frequency of high vulnerabilities (red, orange, and yellow). The zero intervention level includes

all islands less than 140 hectares, of which there are over 530 ecosystem patches (Appendix 1).

That average EVCC1 is the highest here makes sense because so many islands are susceptible to

flooding due to sea level rise. They also receive a high EVCC2a values because often islands

have no “edge” in the edge to core ratio. Intervention level six, productive systems with less

than 10% natural or spontaneous vegetation-type ecosystems received the highest overall

average EVCC. Except for EVCC4 (which makes sense because this is a highly cultivated

ecosystem type that has likely removed native species and decreased biodiversity), all EVCC1,2,3


51

are relatively high. The fact that this study has ranked this intervention level as the second most

vulnerable according to climatic ‘space’ is worth attention. Farming relies upon climatic cycles

that have been learned and adapted to for centuries, and the geographical locations of the patches

of this agricultural ecosystem type are such that climate change projections illustrate some of the

highest changes in temperature and precipitation. Intervention level eight, shrimp and/or salt

production, can mostly be explained by the combination of their close proximity to coastlines

and the almost absent slope of the land. The farming of shrimp and salt requires vast extents of

flat land that are occasionally inundated by sea water at the highest high tides of the year. It is

obvious that sea level rise will greatly impact these areas. Perhaps the most worrisome to

policymakers is that the populated places have received the second highest overall average

vulnerability. It is in the upper half of sea level rise vulnerability and in the lower half of species

sensitivity. The pertinence of its high EVCC2 ranking is questionable because it is not one of the

natural systems, which are more susceptible to ramifications of stochastic processes when they

have irregular shapes or many edges to welcome the perturbations. This human-dominated

system is very buffered by these types of effects. It is also notable that the projected changes in

temperature and precipitation have more of an impact here than other places.

2) Protected areas

The percentage of area protected in each ecosystem type is compared to its mean EVCC value in

(Figure 83). This shows that some of the most vulnerable ecosystem types are also the ones with

the least protection on average. Also, the average EVCC per area (at a resolution of 1 km2) for

the overall EVCC and the four EVCC components was calculated for all areas that are protected,

all areas that are not protected and the whole of Panama (see Table 6). This allows to evaluate


52

how the current protected area network accounts for ecosystem vulnerability to climate change

according to the EVCC. The mean EVCC should be higher within protected areas if the network

generally protects more vulnerable regions. If lower EVCC means that areas outside of the

network are generally more vulnerable to climate change (according to the EVCC) and thus, that

the protected areas network is inadequate in protecting vulnerable areas. We found the latter to

be the case for the overall EVCC, EVCC1 and EVCC3, but the effect is much more pronounced

in EVCC1. Nonetheless, the fact that in almost no case is the average EVCC in protected areas

higher means that the current network of reserve is inadequate to protect the vulnerable areas of

Panama in terms of future climate change. Additionally, something to keep in mind with this

analysis is that protected marine areas are excluded because the ecosystem patch map does not

include aquatic ecosystems.

Table 6: Average EVCCs per area in Panama as a whole

and inside and outside of protected areas

inside

PAs

outside

PAs Panama

EVCC 17.63062 17.787722 17.842452

EVCC1 1.34952 1.978375 2.197648

EVCC2 6.384813 6.094201 5.992869

EVCC3 5.129844 5.132692 5.133686

EVCC4 4.766585 4.582453 4.518249

Comparative Analysis: Protected Areas and EVCC for Ecosystem Type

0

5

10

15

20

25

30

35

7b. B

SOT pantanoso - d

omin. por p

almas

21. V

eg. costera de tra

ns. sobre suelos marinos muy re

c.

7c. B

SOT pantanoso - d

omin. por C

ampnosperm

a

9. M

anglares

6a. B

SOT aluvial ocasionalmente inundado

18a. P

ant. d

e cyperáceas con abund. acum. de mat. o

rg.

7a. B

SOT pantanoso - d

omin. por d

icotile

doneas

13b. S

ist. p

rod. con veg. le

ñosa nat. o

esp. signif. (<

10 %

)

12. P

oblados

6c. B

SOT aluvial ocasionalmente inundado - d

omin. por P

rioria

1b. B

SOT latifo

liado de tie

rras bajas - B

I

1a. B

SOT latifo

liado de tie

rras bajas

1c. B

SOT latifo

liado de tie

rras bajas - P

I

17. V

egetación de páramo

22. A

lbina con escasa vegetación

13a. S

ist. p

rod. con veg. le

ñosa nat. o

esp. signif. (1

0-50 %

)

15. S

ist. p

rod. acuático-camaronera y salina

18b. P

antanos herbáceos salobres

10. B

. deciduo por la

sequia, la

tifoliado de tie

rras bajas

14. P

lantaciones forestales

16. S

abanas arbol. d

e graminoides corto

s inund.

20. F

lujo de lava con escasa vegetación

8b. B

. semideciduo tro

pical de tie

rras bajas - b

astante intervenido

23. C

arriz

ales pant. y

form

. sim. domin. por T

ypha

8c. B

. semideciduo tro

pical de tie

rras bajas - p

oco intervenido

8a. B

. semideciduo tro

pical de tie

rras bajas

6b. B

SOT aluvial ocasionalmente inundado - M

ed. I

3a. B

SOT latifo

liado montano

5. B

SOT latifo

liado nuboso

2b. B

SOT latifo

liado submontano - B

I

3b. B

SOT latifo

liado montano - B

I

4a. B

SOT latifo

liado altim

ontano

2a. B

SOT latifo

liado submontano

2c. B

SOT latifo

liado submontano - P

I

3c. B

SOT latifo

liado montano - P

I

11. Is

la menores de 140 ha

4b. B

SOT latifo

liado altim

ontano - M

ed. I

Ecosystem Type

Average EVCC

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Percentage Protected

Ave_EVCC

Percent Protected

Figure 8 Protected Areas Analysis


54

3) Species richness and endemism

Figure 94: Comparing the overall EVCC with patterns of species richness and endemism in Panama

Species richness and endemicity were calculated for each cell of a 10 x 10 km grid of

Panama (right part of Figure 14). It should be noted that endemism for mammals, birds, and

amphibians is inversely related to the species richness in Panama. That is, there are more birds

than mammals and more mammals than amphibians in the country, but in terms of endemism,

there are the most amphibians and the least birds. For each cell, the overall EVCC was compared

with a map of species richness and endemism to see if any patterns could be extracted or whether

areas of high biodiversity also happened to be more vulnerable to climate change according to

the EVCC (Figure 14). Generally no correlation were found for neither species richness nor

endemism excepting for EVCC3 where regions of high endemism had a low score (this is

vvss..


55

reflected in the overall EVCC as well) (Figure 15). In order to highlight regions that had both

high species richness and overall EVCC, species richness per ecosystem patch was multiplied by

overall EVCC score to yield Map 6. The region that stands out the most in this map is Bocas del

Toro, but the west of Panama, the areas of the Canal and the south of the Darien are highlighted

as well.

Figure 10: Correlation of species richness (a-e) and endemism (f-j) with EVCCs for each cell of a 10x10 km

grid of Panama


56

Map 6 Combining overall EVCC with species richness


57

VI. Conclusions

The objective of this project was to conduct a prelimary assesment of ecosystem vulnerability to

climate change in Panama. The final results include both a set of maps from which spatial

information (including specific information on ecosystem patches and trends) can be extracted

and an analysis of the EVCC in turns of applications relevant to conservation. These were

communicated in a comprehensive format to the host institution as both a hard copy (including a

copy of important GIS data and maps) and a presentation. This project can not only be valuable

for the conservation of biodiversity of Panama both also to other regions by providing a

framework that can be followed to build EVCC assessements in other geographical locations.

The EVCC is a preliminary assessment and is far from being perfect. There are a number

of issues that could be adressed to improve it and a number of uncertainties that need to be

accounted for when applying the EVCC in policy-making. We outline some of them here.

First, certain types of data were unavailable at the time of the assessment but should

preferably be included in future assessments if possible. These include other scenarios of the

CCM than the “business-as-usual” one, higher climate resolution for the CCM and tree

distributions. Moreover, islands are underrated in this assessment because of overlaying issues in

EVCC1 (such that an island’s overall elevation is misrepresented) and because many of them

were not included in EVCC3 since they are so small. Also, future climate for islands is harder to

predict in the future because of micro-climate issues brought by the surrounding water masses.

Finally, the effects of the ranking scheme on the overall EVCC should be investigated. While

efforts were put into using the most relevant scheme for each data type, it might well be that

these had unintended effects on the overall distribution of the EVCC.


58

Second, the EVCC can only be as good as the data it is based on is. There are a number

of uncertainties in the accuracy of the data and thus, the results are not definitive. Possible

problems of accuracies include the digital elevation model of sea level rise vulnerability,

predictions of future climate and species distributions. There are also uncertainties associated

with the theories behind the EVCC. Considerable efforts were put into including the latest

scientific consensys on the effects of climate change on species and ecosystems but these are

always evolving. Likewise, it might well be that the four components selected to build the EVCC

are not enough, or too much, and this should be investigated in light of the most recent scientific

litterature. In general thus, studies like this done in GIS should be used as indicators of where

further research or conservation efforts should be directed. These results cannot replace on-sight

situation analyses, but at least such projects can suggest locations of high concern. It is essential

to use a comprehensive framework to adress uncertainty in data and theory, especially in areas

like climate change where these uncertainties are high.

To conclude, EVCC has the potential to be a very interesthing tool for conservation and

ecology. It is very much a work-in-progress and it has been structured in a flexible way so that it

can be improved easily. One of the biggest challenges to use the EVCC in conservation will be to

decide what its relative importance is compared to other human drivers of change in biodiversity

like habitat loss, over-exploitation and invasive species. Nonewithstanding all the uncertainties

associated with the EVCC, it is the first attempt at qualifying vulnerability to climate change at

the ecosystem level and sets up a template that can be used directly or adjusted according ton

specific needs. Because it estimates an important threat to ecosystems, climate change, which is

usually over-looked in biodiversity assesments, it can become, if properly applied, a very useful

tool for conservation.


59

X. References

Broenniman O, Thuiller W, Hughes G, et al. (2006) Do geographic distribution, niche property

and life form explain plants' vulnerability to global change? Global Change Biology 12,

1079-1093.

Ducklow HW, Baker K, Martinson DG, et al. (2007) Marine pelagic ecosystems: The West

Antarctic Peninsula. Philosophical Transactions of the Royal Society B-Biological

Sciences 362, 67-94.

Frederiksen M, Edwards M, Richardson AJ, Halliday NC, Wanless S (2006) From plankton to

top predators: bottom-up control of a marine food web across four trophic levels. Journal

of Animal Ecology 75, 1259-1268.

Hampe A (2004) Bioclimate envelope models: what they detect and what they hide. Global

Ecology and Biogeography 13, 469-471.

Hawkins BA, Field R, Cornell HV, et al. (2003) Energy, water, and broad-scale geographic

patterns of species richness. Ecology 84, 3105-3117.

Hooper DU, Chapin FS, Ewel JJ, et al. (2005) Effects of biodiversity on ecosystem functioning:

A consensus of current knowledge. Ecological Monographs 75, 3-35.

Ibanez I, Clark JS, Dietze MC, et al. (2006) Predicting biodiversity change: Outside the climate

envelope, beyond the species-area curve. Ecology 87, 1896-1906.

Institute WR (2007) EarthTrends: Environmental Information. Available at

http://earthtrends.wri.org. World Resources Institute, Washington DC.

IPCC (2001) Climate change 2001: impacts, adaptations and vulnerability. Contribution of

Working Group II to the Third Assesment Report of the Intergovernmental Panel on

Climate Change (IPCC). Cambridge University Press, Cambridge.

IPCC (2007) Climate Change 2007: The Physical Science Basis - Summary for Policymakers,

Geneva.

IUCN (2006) 2006 IUCN Red List of Threatened Species. IUCN-SSC

(http://www.iucnredlist.org).

Janzen DH (1983) No Park Is an Island - Increase in Interference from Outside as Park Size

Decreases. Oikos 41, 402-410.

Jiguet F, Julliard R, Thomas CD, et al. (2006) Thermal range predicts bird population resilience

to extreme high temperatures. Ecology Letters 9, 1321-1330.

McGill University (2007) Academic Integrity – Researchers. Available at

http://www.mcgill.ca/integrity/researchers/

Metzger MJ, Leemans R, Schroter D (2005) A multidisciplinary multi-scale framework for

assessing vulnerabilities to global change. International Journal of Applied Earth

Observation and Geoinformation 7, 253-267.

Murcia C (1995) Edge Effects in Fragmented Forests - Implications for Conservation. Trends in

Ecology & Evolution 10, 58-62.

Nature Serve, 2007. InfoNatura website (for species lists and endemism). Available at

www.natureserve.org

Parmesan C (2006) Ecological and evolutionary responses to recent climate change. Annual

Review of Ecology Evolution and Systematics 37, 637-669.

Pearson RG (2006) Climate change and the migration capacity of species. Trends in Ecology &

Evolution 21, 111-113.


60

Peterson G, Allen CR, Holling CS (1998) Ecological resilience, biodiversity, and scale.

Ecosystems 1, 6-18.

Pounds JA, Bustamante MR, Coloma LA, et al. (2006) Widespread amphibian extinctions from

epidemic disease driven by global warming. Nature 439, 161-167.

Sala OE, Chapin FS, Armesto JJ, et al. (2000) Biodiversity - Global biodiversity scenarios for

the year 2100. Science 287, 1770-1774.

Saunders DA, Hobbs RJ, Margules CR (1991) Biological Consequences of Ecosystem

Fragmentation - a Review. Conservation Biology 5, 18-32.

Scholze M, Knorr W, Arnell NW, Prentice IC (2006) A climate-change risk analysis for world

ecosystems. Proceedings of the National Academy of Sciences of the United States of

America 103, 13116-13120.

Schroter D, Cramer W, Leemans R, et al. (2005) Ecosystem service supply and vulnerability to

global change in Europe. Science 310, 1333-1337.

Shafer CL (1999) National park and reserve planning to protect biological diversity: some basic

elements. Landscape and Urban Planning 44, 123-153.

Sutherland WJ (2006) Predicting the ecological consequences of environmental change: a review

of the methods. Journal of Applied Ecology 43, 599-616.

Thomas CD, Cameron A, Green RE, et al. (2004) Extinction risk from climate change. Nature

427, 145-148.

Thuiller W, Lavorel S, Araujo MB (2005a) Niche properties and geographical extent as

predictors of species sensitivity to climate change. Global Ecology and Biogeography 14,

347-357.

Thuiller W, Lavorel S, Araujo MB, Sykes MT, Prentice IC (2005b) Climate change threats to

plant diversity in Europe. Proceedings of the National Academy of Sciences of the United

States of America 102, 8245-8250.

Thuiller W, Midgley GF, Hughes GO, et al. (2006) Endemic species and ecosystem sensitivity to

climate change in Namibia. Global Change Biology 12, 759-776.

Walther GR, Post E, Convey P, et al. (2002) Ecological responses to recent climate change.

Nature 416, 389-395.


61

Appendix 1. Ecosystem Types and Frequencies

UNESCO ecosystem type

# of

patches Total area (ha)

I.A.1.a.(1) Bosque siempreverde ombrofilo tropical latifoliado de tierras bajas 77 1068489.0140

I.A.1.a.(1) Bosque siempreverde ombrofilo tropical latifoliado de tierras bajas -

bastante intervenido 45 654997.4415

I.A.1.a.(1) Bosque siempreverde ombrofilo tropical latifoliado de tierras bajas -

poco intervenido 32 118382.9463

I.A.1.b.(1) Bosque siempreverde ombrofilo tropical latifoliado submontano (500-

1000 m Caribe, 700-1200 m Pacífico) 27 679494.4446


1000 m Caribe, 700-1200 m Pacífico) - bastante intervenido 21 84856.4851


1000 m Caribe, 700-1200 m Pacífico) - poco intervenido 2 4747.5028

I.A.1.c.(1) Bosque siempreverde ombrofilo tropical latifoliado montano (1000-



1500 m Caribe, 1200-1800 m Pacífico) - bastante intervenido 14 45478.7325


1500 m Caribe, 1200-1800 m Pacífico) - poco intervenido 1 5093.5239

I.A.1.d.(1) Bosque siempreverde ombrofilo tropical latifoliado altimontano

(1500-2000 m Caribe, 1800-2300 m Pacífico) 4 38690.5909

I.A.1.d.(1) Bosque siempreverde ombrofilo tropical latifoliado altimontano

(1500-2000 m Caribe, 1800-2300 m Pacífico) - medianamente intervenido 1 6594.3574

I.A.1.e.(1) Bosque siempreverde ombrofilo tropical latifoliado nuboso (2000-


I.A.1.f.(2) Bosque siempreverde ombrofilo tropical aluvial ocasionalmente

inundado 36 77392.5517

I.A.1.f.(2) Bosque siempreverde ombrofilo tropical aluvial ocasionalmente

inundado - medianamente intervenido 1 2795.3689

I.A.1.f.(2)(a) Bosque siempreverde ombrofilo tropical aluvial, ocasionalmente

inundado dominado por Prioria 6 13410.2662

I.A.1.g.(1) Bosque siempreverde ombrofilo tropical pantanoso dominado por

dicotiledóneas 5 4124.8543


palmas 7 3760.6590


Campnosperma 8 23960.6124

I.A.3.a. Bosque semideciduo tropical de tierras bajas 59 708492.0301

I.A.3.a. Bosque semideciduo tropical de tierras bajas - bastante intervenido 76 452877.7101

I.A.3.a. Bosque semideciduo tropical de tierras bajas - poco intervenido 20 59036.6721

I.A.5. Bosque de manglar 137 212650.1390

I.B.1.a.(1) Bosque deciduo por la sequía, latifoliado de tierras bajas 5 7059.7005

Isla menores de 140 hectáreas 530 8291.8601

P. Poblados 21 21170.0424

SP.A. Sistema productivo con vegetación leñosa natural o espontánea

significativa (10-50 %) 60 1739669.8057

SP.B. Sistema productivo con vegetación leñosa natural o espontánea

significativa (<10 %) 46 1142648.6367

SP.B.1. Plantaciones forestales 6 5267.1732

SP.C. Sistema productivo acuático-camaronera y salina 6 7391.1249


62

V.A.2.d. Sabanas arboladas de graminoides cortos inundables 5 4329.8005

V.C.2.b. Vegetación de páramo 2 2484.9664

V.D.1.a. Pantanos de cyperáceas con abundante acumulación de material orgánico 3 1305.0761

V.E.1.a.2. Pantanos herbáceos salobres 4 2002.2752

VI.A.d. Flujo de lava con escasa vegetación 4 5987.8219

VI.B.3. Vegetación costera de transición sobre suelos marinos muy recientes 5 3855.5363

VI.D. Albina con escasa vegetación 1 874.2372

VII.B. Carrizales pantanosos y formaciones similares principalmente de Typha 5 18385.8150

TOTAL 1303 ~7446911


63

Appendix 2. Endemic Species in Panama

Scientific Name Common Name Range (hectares)

Mammalia

Alouatta coibensis Coiba Island Howling Monkey 782260

Bassaricyon pauli Chiriqui Olingo 139384

Bradypus pygmaeus Escudo Island Three-toed Sloth 341

Coendou rothschildi Rothschild’s Porcupine 5829280

Cryptotis endersi Ender's Small-eared Shrew 142297

Cryptotis mera Darien Small-eared Shrew 645692

Dasyprocta coibae Coiban Agouti ?

Dermanura incomitata Isla Escudo Fruit-eating Bat 341

Isthmomys flavidus Yellow Isthmus Rat 647036

Liomys adspersus Panamanian Spiny Pocket Mouse 2636778

Marmosops invictus Slaty Slender Mouse Opossum 2064606

Neacomys pictus Painted Bristly Mouse 322680

Orthogeomys dariensis Darien Pocket Gopher 1832796

Rhipidomys scandens Mount Pirri Climbing Mouse 215628

Tylomys fulviventer A Climbing Rat 10469

Tylomys panamensis Panama Climbing Rat 359135

Aves

Anthracothorax veraguensis Veraguan Mango 3044498

Chlorospingus inornatus Pirre Bush-Tanager 98657

Leptotila battyi Azuero Dove 530265

Margarornis bellulus Beautiful Treerunner 48672

Piculus callopterus Stripe-cheeked Woodpecker 538297

Phylloscartes flavovirens Panama Tyrannulet 1727088

Pselliophorus luteoviridis Yellow-green Finch 121165

Selasphorus ardens Glow-throated Hummingbird 228079

Amphibia

Atelopus certus 10510

Atelopus limosus 44119

Atelopus zeteki Golden Frog 158879

Bolitoglossa anthracina 111536

Bolitoglossa cuna 38846

Bolitoglossa taylori 87142

Bufo peripatetes 12818

Caecilia volcani 700454

Dendrobates arboreus 22822

Dendrobates claudiae 6940

Dendrobates speciosus 37580

Dendrobates vicentei 153146

Eleutherodactylus azueroensis 150366

Eleutherodactylus emcelae 175598

Eleutherodactylus jota 5850


64

Eleutherodactylus laticorpus 28252

Eleutherodactylus monnichorum 13670

Eleutherodactylus museosus 336100

Eleutherodactylus punctariolus 379358

Epipedobates maculates ?

Hyla graceae 187471

Hyla infucata 4965

Hyla thysanota 6550

Oedipina maritime 182

Oscaecilia elongate 5062

Pipa myersi 63184

Rana pipiens Northern Leopard Frog 445030

Rana sp. 2 Common Leopard Frog ?

Scinax altae 1251113

AREA of PANAMA ~7446911

Note: One endemic mammal and two endemic amphibians do not have a range attributed to them

because there are incongruencies between the online database and the Arc files. These

three species have not been recorded in the GIS and therefore cannot be included in the

EVCC4. This is possibly because the online database is more up to date than the Arc

files.


65

Appendix 3. Special Focus Maps

Map 7 Sea Level

Rise Focus: Bocas

del Toro. This

display of one of

the regions in

Panama with some

of the highest

EVCC1 values

shows small

islands, tropical

evergreen swampy

rainforests

dominated by

palms, and

mangroves.

Map 8 Overall

EVCC Focus:

Canal Zone. The

Colón area

frequently has high-

vulnerability

ecosystem patches

of the types:

mangroves and

populated places.


66

Map 9 Overall

EVCC Focus:

Golfo de Chiriquí.

The high-

vulnerability

patches here are

commonly lowland

tropical rainforests,

fairly intensive

agricultural areas,

as well as more

small islands and

mangroves.


67

Appendix 4. Chronogram of Activities & Time Spent

Dates, times, descriptions Total person-

hours spent on

internship

4 January (17:00-19:00)

• Internship Cocktail – introductions and visit to CATHALAC

0

11 & 12 January (8:30-16:30)

• Orientation to CATHALAC and SERVIR with Milton Solano

32

13 January (20:00-20:30)

• Informal meeting with Dr. Roberto Ibáñez to explain preliminary ideas of

the project

1

15 January (19:30-20:30)

• Meeting with supervisor Emil Cherrington to start refining the study

question: “What areas are vulnerable to climate change?”

2

18 & 19 January (8:30-16:30)

• Preliminary requests for data required for the project

• Meeting with Noel Trejos (specialist in Integrated Management of Water

Resources and Community Development) to discuss possible field work

(18 Jan, 13:00)

• Video-meeting with SERVIR Project Manager from NASA Dan Irwin (18

Jan, 15:00)

• Meeting with Joel Peréz Fernandez (specialist in Weather and Climate

Technology) to learn about climatological data and models (19 Jan, 15:00)

32

25 & 26 January (8:30-16:30)

• Preliminary literature review on climate change’s effects on biodiversity

• Overview of spatial analysis, algorithm, and interpolation capabilities of

ArcGIS for temperature data (esp. Krigging)

• Further research on how to measure vulnerability (esp. measuring at

ecoregion vs. ecosystem scale for Panama)

32

1 & 2 February (8:30-16:30)

• Further literature review

o Included 2001 IPCC Report (Vol. 2: Impacts, Adaptation, and

Vulnerability)

• Meeting with Dan Irwin to explain project ideas and to learn what high

resolution climate change models are available (1 Feb, 10:00)

• Teleconference with Jean-Nicolas Poussart in Havana (GIS expert in

assessment and management land-based pollution) to discuss how to make

indices

32


68

5 – 9 February (8:30-16:30) Internship Week #1


• Refined variables to measure ecosystem vulnerability

• Compared UNESCO’s Arc data to its written descriptions on Panama’s

ecosystems

• Informal “Progress Report” presentation to Emil Cherrington: the study

question and in what domains research will be done, goals, objectives,

hypotheses, methods, concerns, and final products (9 Feb, 11:00)

80

12 – 14 February (8:30-16:30) Internship Week #2


• Preparation of Work Plan – Progress Report (due 14 Feb 18:30)

80

1 & 2 March (8:30-16:30)

• Finalizing research and project development

• SERVIR workshop

32

8 & 9 March (8:30-16:30)

• Measuring EVCC variables – determining vulnerability to sea level rise

32

15 March (8:30-16:30)

• Measuring EVCC variables – determining vulnerability to sea level rise

• Preparing informal presentation & mock-presentation to Emil Cherrington

16

16 March

• Informal presentation of projects to other PFSS internship groups

0

17 March (13:00-17:00)

• Measuring EVCC variables – summarizing sea level rise vulnerability &

preparing climatic ‘space’ analysis

4

19 – 22 March (8:30-16:30) Internship Week #3

• Measuring EVCC variables – summarizing sea level rise vulnerability,

analyzing ecosystem geometries, & analyzing climatic ‘space’

• Experts workshop – sat in on an meeting for monitoring climate change

and SERVIR information sessions; described project to Dr. Carey Yeager

of USAID

64

25 March (13:00-17:00)

• Measuring EVCC variables – analyzing ecosystem geometries

4

29 & 30 March (8:30-16:30)

• Measuring EVCC variables – analyzing ecosystem geometries & climatic

‘space’

32

2 & 3 April (13:00-17:00)


‘space’

16

4 April (13:00-19:00)


‘space’

• Information meeting on symposium and final projects, brief meeting with

Dr. Roberto Ibáñez to discuss presentation

12


69

5 & 6 April (9:00-18:00)

• Measuring EVCC variables – analyzing climatic ‘space’ & preparing

species sensitivity analysis

• Measurement of degree of human intervention for ecosystems

• Overlaying EVCC variables with protected areas and degree of human

intervention

36

7 April (10:00-18:00)

• Measuring EVCC variables – analyzing climatic ‘space’ & species

sensitivity

• Overlaying EVCC variables with protected areas and degree of human

intervention

16

16 – 20 April (8:30-16:30) Internship Week #4

• Creating final products (maps, final report)

80

21 & 22 April (10:00-18:00)

• Creating final products

32

23 - 25 April (8:30-17:30)

• Completion of final products

• Presentation of results to CATHALAC

54

• 0

26 April

• Submission of final products

Internship symposium (8:00)

Days spent (not necessarily full): 50 Total person-hours: 721

Note: Most, but not all, days involved both researchers working simultaneously. This is

reflected in the person-hours.


70

Appendix 5. Notes of Gratitude – Contact Information

CATHALAC: Attn: Emil Cherrington, Roxana Segundo

Tel: (507) 317-3200

Fax: (507) 317-3299

E-mail: [email protected], [email protected]

Address: Ciudad del Saber, Edificio 801, Clayton, Panama

Website: http://www.cathalac.org

NASA: Attn: Daniel Irwin

E-mail: [email protected]

Address: Marshall Space Flight Center, Building 4200, Room 120, MSFC,

Huntsville, Alabama 35812

Website: http://www.nasa.gov

Documents

A Preliminary Assessment of Ecosystem Vulnerability to ... · Ecosystem Vulnerability to Climate Change 6 6 DGVM alone or combine it with one of the methods mentioned above (e.g.,